UNIT.City — місце, де люди працюють... КРАЩЕ! Обирай свій простір просто зараз 👉
Валентин ШнайдерMoney
20 February 2026, 15:17
2026-02-20
Mirai Tech receives $10 million in investment and wants to move AI work from servers to smartphones and laptops
Startup Mirai Tech Inc announced that it has raised $10 million in funding to develop infrastructure that will allow AI models to run directly on users' devices, instead of sending requests to remote servers each time.
Startup Mirai Tech Inc announced that it has raised $10 million in funding to develop infrastructure that will allow AI models to run directly on users' devices, instead of sending requests to remote servers each time.
According to a post on LinkedIn by Mirai co-founder Dmytro Shvets, the investment was led by Uncork Capital, and business angels also participated, including David Singleton, Francois Chaubard, Marcin Zukowski, Mati Staniszewski, Gokul Rajaram, Scott «Scooter» Braun, Vijay Krishnan, Ben Parr, Matt Schlicht, Aditya Jami, and others.
The company describes the problem simply: almost every modern smartphone and laptop already has a special AI chip (accelerator), but most developers find it difficult to use it. Usually, separate system engineers, specific optimizations and integrations for different platforms are required. Mirai promises to remove this barrier and make it possible to connect the execution of the model «on the device» in fact as a regular tool — in a few lines of code.
After raising funds, Mirai plans to work with model developers to port them to devices in key scenarios: text, voice, and computer vision. The post also explains the economics: When a model runs on remote servers, each user request means the cost of renting powerful GPUs and maintaining the infrastructure. For real-time products, this quickly becomes expensive, while local execution eliminates the «pay per request» and reduces latency, because the calculations are done on the phone or laptop.
«Cloud computing» in this context means that the model runs not on your device, but somewhere in a data center: the application sends data to the server, the server computes the response, and returns the result. The «on-device» approach does the same thing locally, using built-in AI accelerators, so the service doesn’t need to maintain as much server power for each request.
Previously, dev.ua wrote about how the Ukrainian defense company «The Fourth Law» of entrepreneur Yaroslav Azhnyuk attracted a new round of investment from the American technology group Axon and introduced a new module for interceptor drones.